Apex HV opamp issues.

On a sunny day (Fri, 13 Mar 2015 10:10:38 -0700 (PDT)) it happened George Herold wrote in :

Sometimes {tm) ------||---10k ---- | 1n | |----------||------ | | 100p | 10k ----- 100 k----------------- | | --- - | > ----- --- +

if its oscillatiing you have some pole somewhere. This has worked in the past (don't ask me) but I don't know your loop bandwidth and whatever is in it. Anyways the extra RC (could be some other place that moves) could help against oscillations.

It could work, it could not work, but.. could save time if it does.

OK, I'll give it a try... I was thinking some resistance in series with the 100pF..(?)

(I just ordered some LTC6090's from DK...)

I also noticed that the PA341 has been replaced with the PA441..

How about a resistor in series with the output? Your piezo is probably in the nanofarads.

Cheers

Phil Hobbs

Maybe they were getting lots of complaints! Is it pin-compatible?

try reading this

especially about the different contributions of noise voltage and noise current

Mark

Yeah... more band-aids. I'm not sure it would help. As is the piezo can only react at ~3kHz, so I've already got a bunch of filtering.

I was looking at the noise more closely... turns out it's almost all

(And then some sort of mini pcb to shoe horn it in the the existing pcb.)

George H.

Yup.. I order one of those too.. but only 1.. ~$50 each!

George H.

Thanks Mark, (I've read plenty about noise.) This opamp seems to have a ton of 1/f noise with a 1/f corner freq. that looks to be >10 kHz!

George H.

Apex used to use an AT&T high-voltage CMOS analog process that was astoundingly noisy, 140 nv/rtHz or something crazy like that. Maybe the PA141 was one of those. There was an appnote on compounding that dog with a good low-voltage opamp.

These days, most of their products are regular surfmount parts on PC boards.

You could try my cheap and almost-famous optocoupler-based HV amplifier.

How about the MP118... Spice model on my website, by yours truly ;-)

Model includes .AC and .TRAN ...Jim Thompson

Cheer up, I've measured pHEMTs with 1/f corners of 50 MHz. ;) (It was the otherwise excellent SKY65050.)

Cheers

Phil Hobbs

Thanks but I think I'll try the LTC thing first. The Apex amp sticks out and can be cut off, and a pcb soldered in place.

My only problem is my HV supply is a bit high. right now it runs off ~+155 and -12. (LTC6090 max is 150V) I don't need all the HV ~110-130 would be fine. And low current, I don't know the C of my piezo stack, but max freq is 3kHz. A few mA of current at most. Could I just do a series (beefy) zener? ~20V. With some RC filtering behind if needed.

George H.

Woah this?

That even more $!

George H.

You can use it to launch satellites ;-) ...Jim Thompson

Well it just means filtering is pointless... (in retrospec, I should have measured the noise first)

Replacing the entire opamp with a pcb is something that I can get (some) customers to do in the field. (a TO-3-8 package, eight fat pins and two mounting holes)

George H.

Apex parts definitely make you pay for the privilege of not having to roll your own amp. It's worth it if your volume is low and the cost of failure is high.

At least with some of the Apex amplifiers, they depend on high output capacitance as part of the frequency compensation. Adding (too much) output resistance would make these amplifiers go unstable.

Increasing the load capacitance might be part of reducing closed loop amplifier bandwidth.

HTH-

Yeah, we've been happily paying that cost for years. Now due to their crappy opamp, combined with my lack of due diligence... the return shipping costs to India will make it seem like nothing. (at least I found it now, I think we've only shipped one to India... and you never know we might find someone there who can handle a soldering iron.)

George H.

The MP118, the one I wrote the Spice model for, uses pole-splitting compensation, plus a lead zero... external cap. ...Jim Thompson